NATURAL SELECTION IN MAN 197 
to be impressed with the fact that sentiment has considerably 
influenced opinion on the purely matter of fact problem as to 
whether the infantile death rate is or is not selective. The prob- 
lem of how far selection occurs in the early periods of life is one 
of great difficulty and it is especially important that it be ap- 
proached in an entirely unbiased spirit. In attacking it by statis- 
tical methods it is necessary to be continually on one’s guard 
against falling into the many pitfalls which lie across our path. 
One method by which the problem has been attacked is to 
ascertain the relation between high infant mortality and the 
expectation of life among the survivors. Several investigators 
(von Erben, Bleicher, Gottstein and Rahts) have reported that a 
high infant or child mortality is followed by a relatively low 
mortality in later life. On the other hand, Newsholme in an 
elaborate comparison of the infant and child death rates over 
several districts of England has found that where there is a high 
infant death rate there is also a high death rate of all children up 
to the period of adolescence. Koppe has found a high infant 
mortality correlated with a high death rate in the second year, 
and Prinzing has found a similar correlation between death in the 
first year and deaths from 1 to 4 years of age. Sadayuki’s results 
show that in separate provinces of Germany a high infantile and a 
high child death rate go together. Other investigators (Prinzing, 
v. Vogt, Peiger, Mullhausen) have found (contra Grassl) high 
infant mortality to be correlated with inferiority of recruits 
for military service. 
Those who have concluded from these results, as several have 
done, that the infant death rate cannot be selective have drawn 
an unwarranted inference. Many conditions which produce a 
high infantile death rate are apt to cause a high death rate also in 
childhood and adolescence. Ignorance, poverty, epidemic dis- 
eases and unsanitary surroundings take their toll from people of 
all ages, and the fact that the period beyond infancy is not spared 
because the first year of life is unduly crowded with fatalities, 
in no way proves that the death rate is not selective during the 
whole period. It is not a fair test of the potency of selection to 
